Performance
How to Measure Trading Bot Performance Beyond Win Rate
A bot can win frequently and still lose money. Performance depends on the amount won, the amount lost, the stake used and the payout offered—not just the number of green results. Start with confirmed settlements and keep a clearly defined session or test period.
DTide team · Updated
The four numbers to record first
Return on settled stakes is not a return on your opening account balance. A balance can also change because of deposits, withdrawals, manual trades or another session. Keep those movements separate from the bot’s recorded results.
| Metric | Calculation | What it tells you |
|---|---|---|
| Win rate | Winning settled trades ÷ all settled trades | How often a confirmed result was positive |
| Net profit/loss | Sum of confirmed trade profits and losses | What the recorded trades actually earned or lost |
| Return on settled stakes | Net profit/loss ÷ total settled stake | Net return relative to money staked across completed trades |
| Maximum drawdown | Largest peak-to-trough fall in the confirmed profit curve | The worst decline within that observed sequence |
Example: a 70% win rate that still loses
Suppose ten trades each stake $1. Seven wins earn $0.30 profit each, and three losses lose $1 each. The win rate is 70%, but total profit is $2.10 minus $3.00: a $0.90 loss. Return on the $10 settled stake is −9%.
For a fixed binary quote, break-even probability is stake divided by the total payout on a win. A $1 stake paying $1.30 in total needs about 76.9% wins to break even, before any other applicable costs. Record total payout and net profit separately so they are not confused.
Keep the comparison fair
A favorable historical pattern is a hypothesis to test. Repeatedly picking the best-looking configuration from the same results can overfit that sample. Future outcomes collected after the rule is fixed provide a more useful check. There is no universal trade count that proves profitability for every contract.
- Compare the same contract family, market, direction, duration and barrier or target digit.
- Use one currency and distinguish demo results from real-money results.
- Keep stakes fixed when testing the signal, so stake progression does not hide a weak entry rule.
- Record the chosen settings before the test; change one variable for a separate future comparison.
- Include every accepted trade, with losses and unresolved contracts visible.
Pending trades do not belong in the win-rate denominator
An accepted order with no confirmed broker result is unresolved. Do not count it as a loss because its timer ended, or as a win because the displayed balance increased. Retain the contract reference and reconcile it against the broker’s final status.
DTide’s session overview reports confirmed P/L, return on settled stakes, reserved stake and remaining loss room. Reserved stake represents exposure or queued work; it is not realized profit. Read the scope label before comparing current-session results with restored history.
What to improve when the curve is falling
First check data and execution correctness: the market and direction sent, quote freshness, contract duration, duplicate-order protection and final settlement. A wrong contract or missing result can make the performance report unreliable.
Once the records are trustworthy, evaluate the strategy separately. Compare entry conditions, payout and net return on a new sample. Increasing stakes after losses does not repair an unproven signal, and faster execution alone does not establish a profitable edge.
Common questions
Is a 90% win rate enough?
It depends on the payout and loss size. A contract with frequent small wins can still lose overall if occasional losses outweigh them.
Should pending trades count as losses?
No. Keep them unresolved until a matching final broker result is available. They still represent exposure and may affect whether new orders are allowed.
Do demo profits prove that real trading will be profitable?
No. Demo results test behavior under the observed setup. They do not guarantee future real-money outcomes or remove differences in execution and availability.
Keep learning
Using DTide responsibly
Use this resource alongside the live DTide interface, where current market data and your selected settings are shown. Confirm the market, account type, contract, barrier, duration, stake and session limits before any order is placed.
DTide is independently operated and participates in the Deriv affiliate programme. DTide may earn a commission from an eligible referral without increasing the price paid by that user. DTide is not owned, operated, endorsed or sponsored by Deriv.
Risk reminder: trading derivatives can result in losses. DTide analysis and automation provide context, not guaranteed outcomes.